From 4bf641cb0db1c7a8f2c738659716f9d7fa146210 Mon Sep 17 00:00:00 2001
From: tikikun <daogiatuank54@gmail.com>
Date: Wed, 3 Jan 2024 17:21:24 +0700
Subject: [PATCH] fix high cpu usage

---
 controllers/llamaCPP.h | 1 +
 1 file changed, 1 insertion(+)

diff --git a/controllers/llamaCPP.h b/controllers/llamaCPP.h
index c1017c757..b7b2b27b0 100644
--- a/controllers/llamaCPP.h
+++ b/controllers/llamaCPP.h
@@ -1537,6 +1537,7 @@ struct llama_server_context {
                 "cache\n");
         kv_cache_clear();
       }
+      std::this_thread::sleep_for(std::chrono::milliseconds(5));
       // TODO: Need to implement queueing using CV for better performance
       // std::unique_lock<std::mutex> lock(mutex_tasks);
       // condition_tasks.wait(lock, [&] { return !queue_tasks.empty(); });